Online platform management method and system based on multi-dimensional data AI artificial intelligence

Abstract

The invention discloses an online platform management method and system based on multi-dimensional data AI artificial intelligence, and the method comprises the steps: obtaining a plurality of appearance images of a second-hand mobile phone and user registration information, which are uploaded to an online network platform by a user, and carrying out the image preprocessing of the appearance images, thereby obtaining image data; extracting picture feature information in the image data by adopting a MobileNetv3 network, performing similarity calculation in a preset vector library according to the picture feature information to determine the model of the second-hand mobile phone, and constructing an Internet of Things recovery model according to the user registration information and the model of the second-hand mobile phone, a selling decision model of the second-hand mobile phone is constructed based on the product attention of the user and the Internet of Things recovery model, a transaction order of the user is generated based on the selling decision model and the model of the second-hand mobile phone, and the mobile phone model can be matched through the appearance of the second-hand mobile phone. The attention degree of the user to the product is mined to make an optimal decision so as to match the user to carry out rapid transaction on the online platform, and the reasonable effectiveness of online platform management is improved.

Description

A method and system for managing online platforms based on multidimensional data and AI. Technical Field

[0001] This invention belongs to the field of online transaction platform management technology, and in particular relates to an online platform management method and system based on multi-dimensional data AI artificial intelligence. Background Technology

[0002] In recent years, with the widespread adoption of the internet, an increasing number of online recycling platforms have rapidly emerged. "Internet + recycling" has become the main method for mobile phone recycling. These platforms combine online and offline recycling, breaking geographical limitations and providing convenience for consumers. Smartphones have become an indispensable part of people's lives across all walks of life, serving various functions and uses of applications, and have also become the mainstay of online recycling platforms.

[0003] However, with the increasing demand for smarter and more integrated mobile phones, their functions are also growing. Due to consumers' focus on key product information and the influence of online recycling platforms' reputation on the decisions of various stakeholders in the supply chain, decisions such as recycling and sales prices affect consumers' purchasing and recycling channels for mobile phones—that is, whether they will buy a new or used phone. This makes decision-making difficult, leading to delays in online transactions. Therefore, there is an urgent need to provide an AI-powered online platform management method to address these technical problems. Summary of the Invention

[0004] In view of this, the present invention provides an AI-based online platform management method and system, which can match the appearance of second-hand mobile phones with mobile phone models, deeply explore users' attention to products to make optimal decisions, and facilitate users to conduct quick transactions on the online platform, thereby improving the rationality and effectiveness of online platform management. The specific technical solution adopted is as follows.

[0005] In a first aspect, the present invention provides an online platform management method based on multi-dimensional data AI, comprising the following steps: acquiring multiple appearance images of a used mobile phone uploaded by a user to an online network platform and user registration information, and performing image preprocessing on the appearance images to obtain image data, wherein the image preprocessing includes data cleaning, data augmentation, and image normalization; extracting image feature information from the image data using the MobileNetv3 network, and performing similarity calculation in a preset vector library based on the image feature information to determine the model of the used mobile phone; constructing an IoT recycling model based on the user registration information and the model of the used mobile phone, and constructing a sales decision model for the used mobile phone based on the user's product attention and the IoT recycling model; and generating a user's transaction order based on the sales decision model and the model of the used mobile phone.

[0006] As a preferred embodiment of the above technical solution, the MobileNetv3 network is used to extract image feature information from image data, and similarity calculation is performed in a preset vector library based on the image feature information to determine the model of the used mobile phone. This includes: the MobileNetv3 network comprising a basic convolutional module, an inverse residual module, and a Hard-swish activation function; and the GFL loss function is used to detect the image data, with the corresponding expression being: (1) Among them, This represents the actual value predicted by the model. and This represents two adjacent discretization boundary values, used to discretize the continuous target value y. and These represent the probability values ​​predicted by the model, corresponding to the target value falling within... and The probabilities at the two boundaries; This is an adjustment parameter used to control the averaging of the loss function; the expression for combining the ArcMargin and ResNet algorithms to extract key features of used mobile phones is: (2) Where N represents the number of samples in the batch, This represents the true class label of the i-th sample. This represents the feature vector of the i-th sample and its true class. The angle between the corresponding weight vectors; m is a preset angular interval used to increase the compactness between samples of the same class and the separation between samples of different classes; s is a scaling factor used to amplify the value of logits; This means adding an interval m from the perspective of the true category and calculating the remaining chord values; Let represent the angle between the feature vector of the i-th sample and the weight vector of the j-th class, and n represent the total number of classes.

[0007] As a preferred embodiment of the above technical solution, the Faiss vector database is used to process the vectorized image feature information generated during the feature extraction process, and similarity calculation is performed with all the vectors to be queried in the preset vector library to determine the model of the used mobile phone; the HNSW32 retrieval algorithm and the Faiss vector database are used to quickly match the model of the used mobile phone, and the corresponding expression is: (3) Where A and B represent two vector values ​​for calculating similarity.

[0008] As a preferred embodiment of the above technical solution, an IoT recycling model is constructed based on user registration information and the model of the used mobile phone. A sales decision model for the used mobile phone is then constructed based on user product attention and the IoT recycling model, including: the IoT recycling model incorporates subsidies and sales revenue. Cost recovery Operating costs Fixed infrastructure construction costs and intelligent costs Subsidies and sales revenue include grants to enterprises as rewards and sales revenue from processed equipment, parts, and raw materials. The corresponding expression is: (4) Among them, This indicates that subsidies are being offered to recycling companies to encourage them. This indicates the total number of used mobile phones collected by the recycling company. This indicates the selling price of a used mobile phone. This indicates the number of mobile phones transported from processing center c to the second-hand mobile phone market e. This represents the average weight of each used mobile phone. This means that the parts will be disassembled and sold to parts recyclers. The unit price of the components. Indicates parts recycler The number of mobile phones, This represents the price of the dismantled materials sold to the materials recycling company y. Let represent the number of mobile phones owned by resource recycling company y; recycling cost: includes the cost of purchasing a mobile phone and the recycling fee generated from recycling, based on the cost of on-site collection by the recycler and the cost of the user sending the phone to the recycler via courier. The corresponding expression is: (5) Among them, This represents the average acquisition cost of a used mobile phone. This indicates the proportion of mobile phones transported from each processing center (c) to the material recycling company (y). This represents the average subsidy paid by the recycling company to users who bring their phones to recycling points themselves. This indicates the proportion of door-to-door recycling to the total recycling volume. This represents the average cost incurred by recycling companies using the above recycling methods. This indicates the proportion of total recycling volume that users submit themselves. This represents the cost incurred by the recycling company when a user mails a mobile phone to the recycling center; Operating costs: In the IoT recycling model, users can choose to send their used mobile phones directly to the recycling center via courier. The operating costs for used mobile phones are considered in both scenarios: courier pickup and user self-delivery. (6) Among them, This represents the average operating cost of recycling point h. This represents the average operating cost of recycling center z. This represents the average operating cost of processing center c. , These represent the number of used mobile phones from recycling point h to recycling center z, and the number of used mobile phones mailed by the user to recycling center z, respectively. This represents the number of used mobile phones from recycling center z to processing center c. This represents the unit cost of manual management of mobile phones based on the Internet of Things. This represents the minimum number of alternative recycling centers z to be constructed. This represents the maximum number of alternative recycling centers that can be built. This represents the minimum number of candidate centers to be established, This represents the maximum number of alternative processing centers (c) to be built; Fixed infrastructure construction cost: the construction cost of the recycling center and the processing center, the corresponding expression is: (7) Among them, This represents the cost required to construct recycling point h. This represents the cost required to build the recycling center z. , All variables are 0 to 1; Transportation costs: In the entire logistics network for used mobile phone recycling, the series of transportation-related costs incurred from various upstream nodes to transport used mobile phones are represented by the following expression: (8) Among them, This represents the unit transportation cost from recycling point h to recycling center z. This represents the unit transportation cost from processing center c to secondhand market e. This indicates the process center C leads to the parts recycler. The unit transportation cost This indicates the process center C leads to the parts recycler. The unit transportation cost This represents the unit transportation cost from processing center c to waste treatment plant f. This represents the distance between the recycling point h and the recycling center z. This represents the distance between recycling center c and processing center c. This represents the distance between processing center c and secondhand market e. This indicates that processing center C and parts recyclers The distance between them This represents the distance between processing center c and raw material recycling company y. This represents the distance between processing center c and waste treatment plant f; Smart cost: includes IoT technology operation costs and system maintenance costs, the corresponding expression is: (9) Among them, This indicates the number of times an IoT device undergoes system maintenance per year. This indicates the cost of each maintenance session for IoT devices. This represents the average operating cost of recycling point h. This represents the average operating cost of recycling center z. The objective function represents the average operating cost of processing center c; it consists of: .

[0009] As a preferred embodiment of the above technical solution, when the utility of a user purchasing a new mobile phone is greater than the utility gained from purchasing a used mobile phone, Users purchase new mobile phones from retailers on online platforms; when the utility of buying a used mobile phone is greater than the utility of buying a new mobile phone, Users purchase used mobile phones from self-operated recycling platforms on online platforms, increasing the market demand for new mobile phones. The market demand for second-hand mobile phones ;in, This represents the user's utility function for the new phone. This represents the user's utility function for used mobile phones. This indicates the price at which retailers sell new mobile phones under a self-operated strategy. This indicates the price for reselling a used mobile phone. This indicates users' acceptance of second-hand mobile phones. This indicates the level of reputation and perception of the recycling platform among users. This indicates the level of attention users pay to product information.

[0010] As a preferred embodiment of the above technical solution, the retailer's profit includes the profit from new mobile phones, and the recycling platform's profit includes the revenue from selling used mobile phones and transferring them to processors, minus the cost of paying the recycling price, product promotion costs, and the cost of building its own platform. The retailer's profit function and the recycling platform's profit function are as follows: (10) (11) Based on the retailer being the leader of the supply chain, the supply chain members engage in a two-stage Stackelberg game: In the first stage, the retailer determines the optimal retail price of the new mobile phone. In the second phase, the online recycling platform determines the optimal recycling price. Product promotion investment level and the selling price of used mobile phones Under the strategy of self-operated sales of used mobile phones by recycling platforms, when the following conditions are met... and At that time, retailers and recycling platforms make optimal decisions; the optimal retail price for a new mobile phone is: Best used mobile phone retail price: Optimal recycling price: Optimal level of investment in product promotion: ;in, Indicates the potential size of the market for recycling. This indicates the user's sensitivity to recycling prices. 'b' represents the proportion of low-quality used mobile phones transferred to processors, 'k' represents the transfer price to processors, and 'k' represents the product promotion cost coefficient. This indicates users' acceptance of second-hand mobile phones, and F represents the cost of building and operating a recycling platform.

[0011] As a preferred technical solution, the recycling platform entrusts refurbished and processed second-hand mobile phones to retailers for sale. Users will see the recycling platform's origin and verification information for the second-hand mobile phones on the retailer's platform interface. When the user's utility from buying a new mobile phone is greater than the utility from buying a second-hand mobile phone... Users can buy new phones from retailers; when the utility of buying a used phone is greater than the utility of buying a new phone, the market demand for new phones will... When the utility of buying a used mobile phone is greater than the utility of buying a new mobile phone, Users buy used mobile phones on self-operated recycling platforms, increasing the market demand for new mobile phones. Second-hand mobile phones; among them, the market demand for second-hand mobile phones Retailer profits include profits from new phones and commissions from recycling platforms, while recycling platforms' profits include revenue from the sale of used phones and transfers to processors, minus input and output payments. (12) (13) Based on the retailer as the leader of the supply chain, the supply chain members conduct a two-stage Stackelberg game: In the first stage, the retailer determines the optimal retail price of the new mobile phone. In the second phase, the online recycling platform determines the optimal recycling price. Product promotion investment level and the selling price of used mobile phones ;in, 'f' represents the amount recycled, and 'f' represents the commission rate that the recycling platform needs to pay.

[0012] As a preferred technical solution, under the strategy of self-operated sales of used mobile phones by the recycling platform, when the following conditions are met... and At that time, retailers and recycling platforms have an optimal decision: the optimal retail price of a new mobile phone. Best retail price for used mobile phones Optimal recycling price: Optimal level of investment in product promotion: .

[0013] As a preferred embodiment of the above technical solution, generating a user's transaction order based on the sales decision model and the model of the used mobile phone includes: obtaining an optimal decision by performing sensitivity analysis based on the user's level of attention to product information and the reputation of the recycling platform, wherein the optimal decision includes the optimal used mobile phone sales price and the optimal new mobile phone retail price; and generating a user's transaction order based on the optimal used mobile phone sales price and the model of the used mobile phone.

[0014] Secondly, the present invention also provides an online platform management system based on multidimensional data AI, applied to the aforementioned online platform management method based on multidimensional data AI, comprising: a data acquisition unit, used to acquire multiple appearance images of a used mobile phone uploaded by a user to an online network platform and user registration information, and to perform image preprocessing on the appearance images to obtain image data, wherein the image preprocessing includes data cleaning, data augmentation, and image normalization; a feature extraction unit, used to extract image feature information from the image data using a MobileNetv3 network, and to perform similarity calculation in a preset vector library based on the image feature information to determine the model of the used mobile phone; a model building unit, used to build an IoT recycling model based on the user registration information and the model of the used mobile phone, and to build a sales decision model for the used mobile phone based on the user's product attention and the IoT recycling model; and an order generation unit, used to generate a transaction order for the user based on the sales decision model and the model of the used mobile phone.

[0015] This invention provides an online platform management method and system based on multi-dimensional data AI. It acquires multiple images of a used mobile phone uploaded by a user to an online platform, along with the user's registration information. The images are preprocessed to obtain image data. MobileNetv3 is used to extract image feature information from the image data. Similarity calculations are performed in a preset vector library based on these image feature information to determine the model of the used mobile phone. An IoT recycling model is constructed based on the user's product attention and the IoT recycling model. A sales decision model for the used mobile phone is then built based on the user's product attention and the IoT recycling model. Finally, a transaction order is generated based on the sales decision model and the model of the used mobile phone. By matching the appearance of the used mobile phone with the model, and deeply mining the user's product attention, optimal decisions are made to facilitate rapid transactions on the online platform, improving the rationality and effectiveness of online platform management. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 is a flowchart of the online platform management method based on multidimensional data AI provided by the present invention; Figure 2 is a structural block diagram of the online platform management system based on multidimensional data AI provided by the present invention. Detailed Implementation

[0018] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0019] Referring to Figure 1, this invention provides an online platform management method based on multi-dimensional data AI, comprising the following steps: S1: acquiring multiple appearance images of a used mobile phone uploaded by a user to an online network platform and user registration information, and performing image preprocessing on the appearance images to obtain image data, wherein the image preprocessing includes data cleaning, data augmentation, and image normalization; S2: extracting image feature information from the image data using the MobileNetv3 network, and performing similarity calculation in a preset vector library based on the image feature information to determine the model of the used mobile phone; S3: constructing an IoT recycling model based on the user registration information and the model of the used mobile phone, and constructing a sales decision model for the used mobile phone based on the user's product attention and the IoT recycling model; S4: generating a user's transaction order based on the sales decision model and the model of the used mobile phone.

[0020] In this embodiment, the MobileNetv3 network is used to extract image feature information from image data, and similarity calculation is performed in a preset vector library based on the image feature information to determine the model of the used mobile phone. This includes: the MobileNetv3 network comprising a basic convolutional module, an inverted residual module, and a Hard-swish activation function; and the GFL loss function is used to detect the image data. The corresponding expression is: (1) Among them, This represents the actual value predicted by the model. and This represents two adjacent discretization boundary values, used to discretize the continuous target value y. and These represent the probability values ​​predicted by the model, corresponding to the target value falling within... and The probabilities at the two boundaries; This is an adjustment parameter used to control the averaging of the loss function; the expression for combining the ArcMargin and ResNet algorithms to extract key features of used mobile phones is: (2) Where N represents the number of samples in the batch, This represents the true class label of the i-th sample. This represents the feature vector of the i-th sample and its true class. The angle between the corresponding weight vectors; m is a preset angular interval used to increase the compactness between samples of the same class and the separation between samples of different classes; s is a scaling factor used to amplify the value of logits; This means adding an interval m from the perspective of the true category and calculating the remaining chord values; Let represent the angle between the feature vector of the i-th sample and the weight vector of the j-th class, and n represent the total number of classes.

[0021] It's worth noting that the GFL loss function removes the Centerness branch from the detection head. This improvement saves the model from a large number of convolution operations on this branch during training and inference, effectively reducing the computational overhead of the detection head. This reduced computational resource requirement allows the model to process images faster while maintaining detection accuracy, further improving detection efficiency. The feature extraction module, based on deep feature learning, introduces metric learning methods to implement the application. Building upon the accurate localization of discarded mobile phones through subject detection, this module uses deep feature learning to mine the content features of the target subject (second-hand mobile phone), and simultaneously uses metric learning methods to measure and analyze these features, thereby deeply mining more comprehensive and accurate feature information of the target subject (second-hand mobile phone). These features serve as crucial evidence for subsequent mobile phone signal recognition and classification, possessing extremely high value. By extracting representative features, different mobile phone models can be effectively distinguished at the feature level in a high-dimensional feature space, providing a solid data foundation for subsequent recognition and classification tasks. ResNet50 plays a crucial role in feature extraction from detected secondhand mobile phone images. It can perform detailed analysis from multiple dimensions such as the phone's appearance, color, and logos to extract high-dimensional feature vectors.

[0022] Among them, ResNet50 demonstrates significant advantages, possessing powerful feature representation capabilities. It can accurately capture subtle features of different mobile phone brands, such as camera arrangement, brand logo position and style, and transform these features into corresponding feature vectors, providing a solid data foundation for subsequent classification and matching tasks. ResNet50 is pre-trained on large-scale datasets, and its learned general features exhibit excellent generalization performance. When faced with images of discarded mobile phones in different scenarios, ResNet50 can still stably extract effective features, which greatly improves the accuracy and reliability of vector retrieval, thereby effectively enhancing the efficiency and quality of the entire image analysis process. Users can upload images of their own or others' used or new mobile phones for recycling or purchase according to their needs.

[0023] It should be understood that by acquiring multiple images of the appearance of a used mobile phone uploaded by a user to an online network platform, along with the user's registration information, and performing image preprocessing on the images to obtain image data, the MobileNetv3 network is used to extract image feature information from the image data. Based on the image feature information, similarity calculations are performed in a preset vector library to determine the model of the used mobile phone. An IoT recycling model is constructed based on the user's product attention and the IoT recycling model. A sales decision model for the used mobile phone is then constructed based on the user's product attention and the IoT recycling model. Based on the sales decision model and the model of the used mobile phone, a user's transaction order is generated. By matching the appearance of the used mobile phone with the mobile phone model, and by deeply mining the user's product attention, optimal decisions are made to facilitate rapid transactions on the online platform, thereby improving the rationality and effectiveness of online platform management.

[0024] Optionally, the Faiss vector database is used to process the vectorized image feature information generated during the feature extraction process, and similarity calculation is performed with all query vectors in a preset vector library to determine the model of the used mobile phone; the HNSW32 retrieval algorithm and the Faiss vector database are used to quickly match the model of the used mobile phone, and the corresponding expression is: (3) Where A and B represent two vector values ​​for calculating similarity.

[0025] In this embodiment, an IoT recycling model is constructed based on user registration information and the model of the used mobile phone. A sales decision model for the used mobile phone is then constructed based on the user's product attention and the IoT recycling model, including: the IoT recycling model includes subsidies and sales revenue. Cost recovery Operating costs Fixed infrastructure construction costs and intelligent costs Subsidies and sales revenue include grants to enterprises as rewards and sales revenue from processed equipment, parts, and raw materials. The corresponding expression is: (4) Among them, This indicates that subsidies are being offered to recycling companies to encourage them. This indicates the total number of used mobile phones collected by the recycling company. This indicates the selling price of a used mobile phone. This indicates the number of mobile phones transported from processing center c to the second-hand mobile phone market e. This represents the average weight of each used mobile phone. This means that the parts will be disassembled and sold to parts recyclers. The unit price of the components. Indicates parts recycler The number of mobile phones, This represents the price of the dismantled materials sold to the materials recycling company y. Let represent the number of mobile phones owned by resource recycling company y; recycling cost: includes the cost of purchasing a mobile phone and the recycling fee generated from recycling, based on the cost of on-site collection by the recycler and the cost of the user sending the phone to the recycler via courier. The corresponding expression is: (5) Among them, This represents the average acquisition cost of a used mobile phone. This indicates the proportion of mobile phones transported from each processing center (c) to the material recycling company (y). This represents the average subsidy paid by the recycling company to users who bring their phones to recycling points themselves. This indicates the proportion of door-to-door recycling to the total recycling volume. This represents the average cost incurred by recycling companies using the above recycling methods. This indicates the proportion of total recycling volume that users submit themselves. This represents the cost incurred by the recycling company when a user mails a mobile phone to the recycling center; Operating costs: In the IoT recycling model, users can choose to send their used mobile phones directly to the recycling center via courier. The operating costs for used mobile phones are considered in both scenarios: courier pickup and user self-delivery. (6) Among them, This represents the average operating cost of recycling point h. This represents the average operating cost of recycling center z. This represents the average operating cost of processing center c. , These represent the number of used mobile phones from recycling point h to recycling center z, and the number of used mobile phones mailed by the user to recycling center z, respectively. This represents the number of used mobile phones from recycling center z to processing center c. This represents the unit cost of manual management of mobile phones based on the Internet of Things. This represents the minimum number of alternative recycling centers z to be constructed. This represents the maximum number of alternative recycling centers that can be built. This represents the minimum number of candidate centers to be established, This represents the maximum number of alternative processing centers (c) to be built; Fixed infrastructure construction cost: the construction cost of the recycling center and the processing center, the corresponding expression is: (7) Among them, This represents the cost required to construct recycling point h. This represents the cost required to build the recycling center z. , All variables are 0 to 1; Transportation costs: In the entire logistics network for used mobile phone recycling, the series of transportation-related costs incurred from various upstream nodes to transport used mobile phones are represented by the following expression: (8) Among them, This represents the unit transportation cost from recycling point h to recycling center z. This represents the unit transportation cost from processing center c to secondhand market e. This indicates the process center C leads to the parts recycler. The unit transportation cost This indicates the process center C leads to the parts recycler. The unit transportation cost This represents the unit transportation cost from processing center c to waste treatment plant f. This represents the distance between the recycling point h and the recycling center z. This represents the distance between recycling center c and processing center c. This represents the distance between processing center c and secondhand market e. This indicates that processing center C and parts recyclers The distance between them This represents the distance between processing center c and raw material recycling company y. This represents the distance between processing center c and waste treatment plant f; Smart cost: includes IoT technology operation costs and system maintenance costs, the corresponding expression is: (9) Among them, This indicates the number of times an IoT device undergoes system maintenance per year. This indicates the cost of each maintenance session for IoT devices. This represents the average operating cost of recycling point h. This represents the average operating cost of recycling center z. The objective function represents the average operating cost of processing center c; it consists of: .

[0026] It should be noted that when the utility a user gains from buying a new phone is greater than the utility gained from buying a used phone, Users purchase new mobile phones from retailers on online platforms; when the utility of buying a used mobile phone is greater than the utility of buying a new mobile phone, Users purchase used mobile phones from self-operated recycling platforms on online platforms, increasing the market demand for new mobile phones. The market demand for second-hand mobile phones ;in, This represents the user's utility function for the new phone. This represents the user's utility function for used mobile phones. This indicates the price at which retailers sell new mobile phones under a self-operated strategy. This indicates the price for reselling a used mobile phone. This indicates users' acceptance of second-hand mobile phones. This indicates the level of reputation and perception of the recycling platform among users. This indicates the level of attention users pay to product information.

[0027] This invention's online platform comprises a supply chain consisting of retailers, an internet recycling platform, and consumers. Retailers are responsible for selling new mobile phones, while the recycling platform is responsible for collecting used mobile phones from consumers. Simultaneously, it invests technology and resources in promoting used mobile phones, classifying and processing the collected phones, and refurbishing high-quality used phones. Notably, under a self-operated strategy, the recycling platform also builds its own platform to sell used mobile phones, allowing users to purchase them through its interface. Under a commissioned strategy, the recycling platform commissions retailers to sell used mobile phones, paying them a commission. Users can see the recycling platform's source and verification information on the retailer's interface before making a purchase.

[0028] Optionally, the retailer's profit includes the profit from new phones, and the recycling platform's profit includes the revenue from selling used phones and transferring them to processors, minus the cost of paying the recycling price, product promotion costs, and the cost of building its own platform. The retailer's profit function and the recycling platform's profit function are as follows: (10) (11) Based on the retailer being the leader of the supply chain, the supply chain members engage in a two-stage Stackelberg game: In the first stage, the retailer determines the optimal retail price of the new mobile phone. In the second phase, the online recycling platform determines the optimal recycling price. Product promotion investment level and the selling price of used mobile phones Under the strategy of self-operated sales of used mobile phones by recycling platforms, when the following conditions are met... and At that time, retailers and recycling platforms make optimal decisions; the optimal retail price for a new mobile phone is: Best used mobile phone retail price: Optimal recycling price: Optimal level of investment in product promotion: ;in, Indicates the potential size of the market for recycling. This indicates the user's sensitivity to recycling prices. 'b' represents the proportion of low-quality used mobile phones transferred to processors, 'k' represents the transfer price to processors, and 'k' represents the product promotion cost coefficient. This indicates users' acceptance of second-hand mobile phones, and F represents the cost of building and operating a recycling platform.

[0029] In this embodiment, the recycling platform entrusts the refurbished used mobile phones to retailers for sale. Users will see the source of the used mobile phones from the recycling platform and the verification information on the retailer's platform interface. When the user's utility from buying a new mobile phone is greater than the utility from buying a used mobile phone... Users can buy new phones from retailers; when the utility of buying a used phone is greater than the utility of buying a new phone, the market demand for new phones will... When the utility of buying a used mobile phone is greater than the utility of buying a new mobile phone, Users buy used mobile phones on self-operated recycling platforms, increasing the market demand for new mobile phones. Second-hand mobile phones; among them, the market demand for second-hand mobile phones Retailer profits include profits from new phones and commissions from recycling platforms, while recycling platforms' profits include revenue from the sale of used phones and transfers to processors, minus input and output payments. (12) (13) Based on the retailer as the leader of the supply chain, the supply chain members conduct a two-stage Stackelberg game: In the first stage, the retailer determines the optimal retail price of the new mobile phone. In the second phase, the online recycling platform determines the optimal recycling price. Product promotion investment level and the selling price of used mobile phones ;in, 'f' represents the amount recycled, and 'f' represents the commission rate that the recycling platform needs to pay.

[0030] It should be noted that under the strategy of the recycling platform selling used mobile phones directly, when the following conditions are met... and At that time, retailers and recycling platforms have an optimal decision: the optimal retail price of a new mobile phone. Best retail price for used mobile phones Optimal recycling price: Optimal level of investment in product promotion: .

[0031] The process of generating a user's transaction order based on the sales decision model and the model of the used mobile phone includes: obtaining the optimal decision by conducting sensitivity analysis based on the user's level of attention to product information and the reputation of the recycling platform, wherein the optimal decision includes the optimal used mobile phone sales price and the optimal new mobile phone retail price; and generating a user's transaction order based on the optimal used mobile phone sales price and the model of the used mobile phone.

[0032] Specifically, as consumers (users) pay more attention to product information, the optimal retail price of new mobile phones will rise, the optimal trade-in price will fall, and the optimal level of promotional investment will increase. Consumers will have less preference for second-hand mobile phones due to concerns about data leakage, and will have more favor and trust for new mobile phones. Therefore, the increased demand for new mobile phones will give them a market advantage, and retailers will correspondingly increase the retail price of new mobile phones. In addition, increased attention to product information indicates that consumers are more sensitive to the performance of their own mobile phones. Therefore, consumers are unwilling to trade in their phones due to distrust of data security during trade-in. Trade-in platforms will increase investment in product promotion to gain consumer trust, thereby increasing consumers' willingness to trade in their phones. At the same time, trade-in platforms will lower trade-in prices to maintain a certain level of profit.

[0033] When product information attention is below a certain threshold, optimal used phone sales will increase as product information attention increases. At this point, consumers are relatively less price-sensitive to product information, and both recycling volume and product promotion levels are increasing, leading to higher recycling costs. Recycling platforms will gradually raise used phone prices to generate more revenue. Conversely, when product information attention exceeds a certain threshold, optimal used phone prices will decrease as product information attention increases. Consumers become more sensitive to product information (configuration parameters or performance), and price used phones to compete in the market with lower prices. Therefore, it can be concluded that the peak price of used phones occurs at a certain threshold of product information attention.

[0034] Furthermore, as consumers pay more attention to product information, the optimal retail price of new mobile phones and the optimal level of promotional investment will increase, while the optimal trade-in price of used mobile phones will decrease. At this point, consumers' preference for and acceptance of used mobile phones decreases, giving new mobile phones a greater market advantage. Therefore, the retail price of new mobile phones will increase accordingly. Consumers' increased product attention makes them less willing to trade in their phones. Simply increasing economic incentives will raise trade-in costs and ultimately harm the profits of the trade-in platform. In this situation, the trade-in price will decrease, and increased consumer product attention will further hinder their trade-in behavior, leading the trade-in platform to increase its investment in product promotion costs. When product information attention is below a certain threshold, and consumers are not so sensitive to product specifications, the trade-in platform will gradually increase the price of used mobile phones to gain more revenue. When product information attention is above a certain threshold, the trade-in platform will decrease the price of used mobile phones as consumers become more concerned about product performance.

[0035] Referring to Figure 2, the present invention also provides an online platform management system based on multidimensional data AI, applied to the aforementioned online platform management method based on multidimensional data AI, comprising: a data acquisition unit, used to acquire multiple appearance images of a used mobile phone uploaded by a user to an online network platform and user registration information, and to perform image preprocessing on the appearance images to obtain image data, wherein the image preprocessing includes data cleaning, data augmentation, and image normalization; a feature extraction unit, used to extract image feature information from the image data using a MobileNetv3 network, and to perform similarity calculation in a preset vector library based on the image feature information to determine the model of the used mobile phone; a model building unit, used to build an IoT recycling model based on the user registration information and the model of the used mobile phone, and to build a sales decision model for the used mobile phone based on the user's product attention and the IoT recycling model; and an order generation unit, used to generate a transaction order for the user based on the sales decision model and the model of the used mobile phone.

[0036] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limitations; therefore, other examples of exemplary embodiments may have different values.

[0037] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0038] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for managing an online platform based on multidimensional data AI, characterized in that, The process includes the following steps: acquiring multiple images of the appearance of a used mobile phone uploaded by a user to an online platform, along with the user's registration information; performing image preprocessing on the images to obtain image data, including data cleaning, data augmentation, and image normalization; extracting image feature information from the image data using the MobileNetv3 network, and calculating similarity in a preset vector library based on the image feature information to determine the model of the used mobile phone; constructing an IoT recycling model based on the user's registration information and the model of the used mobile phone, and constructing a sales decision model for the used mobile phone based on the user's product attention and the IoT recycling model; and generating a transaction order for the user based on the sales decision model and the model of the used mobile phone.

2. The online platform management method based on multi-dimensional data AI artificial intelligence according to claim 1, characterized in that, The MobileNetv3 network is used to extract image feature information from image data, and similarity calculation is performed in a preset vector library based on the image feature information to determine the model of the used mobile phone. This includes: the MobileNetv3 network comprising a basic convolutional module, an inverse residual module, and a Hard-swish activation function; and the GFL loss function is used for image data detection. The corresponding expression is: (1) Among them, This represents the actual value predicted by the model. and This represents two adjacent discretization boundary values, used to discretize the continuous target value y. and These represent the probability values ​​predicted by the model, corresponding to the target value falling within... and The probabilities at the two boundaries; This is an adjustment parameter used to control the averaging of the loss function; the expression for combining the ArcMargin and ResNet algorithms to extract key features of used mobile phones is: (2) Where N represents the number of samples in the batch. This represents the true class label of the i-th sample. This represents the feature vector of the i-th sample and its true class. The angle between the corresponding weight vectors; m is a preset angular interval used to increase the compactness between samples of the same class and the separation between samples of different classes; s is a scaling factor used to amplify the value of logits; This means adding an interval m from the perspective of the true category and calculating the remaining chord values; Let represent the angle between the feature vector of the i-th sample and the weight vector of the j-th class, and n represent the total number of classes.

3. The online platform management method based on multi-dimensional data AI artificial intelligence according to claim 2, characterized in that, Also includes: The Faiss vector database is used to process the vectorized image feature information generated during feature extraction, and similarity calculations are performed with all query vectors in a pre-defined vector library to determine the model of the used mobile phone. The HNSW32 retrieval algorithm and the Faiss vector database are used for fast matching of the used mobile phone model; the corresponding expression is: (3) Where A and B represent two vector values ​​for calculating similarity.

4. The online platform management method based on multi-dimensional data AI artificial intelligence according to claim 1, characterized in that, An IoT recycling model is constructed based on user registration information and the model of the used mobile phone. A sales decision model for the used mobile phone is then built based on user product interest and the IoT recycling model, including: the IoT recycling model incorporates subsidies and sales revenue. Cost recovery Operating costs Fixed infrastructure construction costs and intelligent costs Subsidies and sales revenue include grants to enterprises as rewards and sales revenue from processed equipment, parts, and raw materials. The corresponding expression is: (4) Among them, This indicates that subsidies are being offered to recycling companies to encourage them. This indicates the total number of used mobile phones collected by the recycling company. This indicates the selling price of a used mobile phone. This indicates the number of mobile phones transported from processing center c to the second-hand mobile phone market e. This represents the average weight of each used mobile phone. This means that the parts will be disassembled and sold to parts recyclers. The unit price of the components. Indicates parts recycler The number of mobile phones, This represents the price of the dismantled materials sold to the materials recycling company y. Let represent the number of mobile phones owned by resource recycling company y; recycling cost: includes the cost of purchasing a mobile phone and the recycling fee generated from recycling, based on the cost of on-site collection by the recycler and the cost of the user sending the phone to the recycler via courier. The corresponding expression is: (5) Among them, This represents the average acquisition cost of a used mobile phone. This indicates the proportion of mobile phones transported from each processing center (c) to the material recycling company (y). This represents the average subsidy paid by the recycling company to users who bring their phones to recycling points themselves. This indicates the proportion of door-to-door recycling to the total recycling volume. This represents the average cost incurred by recycling companies using the above recycling methods. This indicates the proportion of total recycling volume that users submit themselves. This represents the cost incurred by the recycling company when a user mails a mobile phone to the recycling center; Operating costs: In the IoT recycling model, users can choose to send their used mobile phones directly to the recycling center via courier. The operating costs for used mobile phones are considered in both scenarios: courier pickup and user self-delivery. (6) Among them, This represents the average operating cost of recycling point h. This represents the average operating cost of recycling center z. This represents the average operating cost of processing center c. 、 These represent the number of used mobile phones from recycling point h to recycling center z, and the number of used mobile phones mailed by the user to recycling center z, respectively. This represents the number of used mobile phones from recycling center z to processing center c. This represents the unit cost of manual management of mobile phones based on the Internet of Things. This represents the minimum number of alternative recycling centers z to be constructed. This represents the maximum number of alternative recycling centers that can be built. This represents the minimum number of candidate centers to be established, This represents the maximum number of alternative processing centers (c) to be built; Fixed infrastructure construction cost: the construction cost of the recycling center and the processing center, the corresponding expression is: (7) Among them, This represents the cost required to construct recycling point h. This represents the cost required to build the recycling center z. 、 All variables are 0 to 1; Transportation costs: In the entire logistics network for used mobile phone recycling, the series of transportation-related costs incurred from various upstream nodes to transport used mobile phones are represented by the following expression: (8) Among them, This represents the unit transportation cost from recycling point h to recycling center z. This represents the unit transportation cost from processing center c to secondhand market e. This indicates the process center C leads to the parts recycler. The unit transportation cost This indicates the process center C leads to the parts recycler. The unit transportation cost This represents the unit transportation cost from processing center c to waste treatment plant f. This represents the distance between the recycling point h and the recycling center z. This represents the distance between recycling center c and processing center c. This represents the distance between processing center c and secondhand market e. This indicates that processing center C and parts recyclers The distance between them This represents the distance between processing center c and raw material recycling company y. This represents the distance between processing center c and waste treatment plant f; Smart cost: includes IoT technology operation costs and system maintenance costs, the corresponding expression is: (9) Among them, This indicates the number of times an IoT device undergoes system maintenance per year. This indicates the cost of each maintenance session for IoT devices. This represents the average operating cost of recycling point h. This represents the average operating cost of recycling center z. The objective function represents the average operating cost of processing center c; it consists of: 。 5. The online platform management method based on multi-dimensional data AI artificial intelligence according to claim 4, characterized in that, Also includes: When the utility a user gains from buying a new phone is greater than the utility gained from buying a used phone. Users purchase new mobile phones from retailers on online platforms; when the utility of buying a used mobile phone is greater than the utility of buying a new mobile phone, Users purchase used mobile phones from self-operated recycling platforms on online platforms, increasing the market demand for new mobile phones. The market demand for second-hand mobile phones ;in, This represents the user's utility function for the new phone. This represents the user's utility function for used mobile phones. This indicates the price at which retailers sell new mobile phones under a self-operated strategy. This indicates the price for reselling a used mobile phone. This indicates users' acceptance of second-hand mobile phones. This indicates the level of reputation and perception of the recycling platform among users. This indicates the level of attention users pay to product information.

6. The online platform management method based on multi-dimensional data AI artificial intelligence according to claim 5, characterized in that, Retailers' profits include profits from new phones, while recycling platforms' profits include revenue from selling used phones and transferring them to processors, minus the costs of paying recycling fees and building their own platforms. The profit functions for retailers and recycling platforms are as follows: （10） (11) Based on the retailer being the leader of the supply chain, the supply chain members engage in a two-stage Stackelberg game: In the first stage, the retailer determines the optimal retail price of the new mobile phone. In the second phase, the online recycling platform determines the optimal recycling price. Product promotion investment level and the selling price of used mobile phones Under the strategy of self-operated sales of used mobile phones by recycling platforms, when the following conditions are met... and At that time, retailers and recycling platforms make optimal decisions; the optimal retail price for a new mobile phone is: Best used mobile phone retail price: Optimal recycling price: Optimal level of investment in product promotion: ;in, Indicates the potential size of the market for recycling. This indicates the user's sensitivity to recycling prices. 'b' represents the proportion of low-quality used mobile phones transferred to processors, 'k' represents the transfer price to processors, and 'k' represents the product promotion cost coefficient. This indicates users' acceptance of second-hand mobile phones, and F represents the cost of building and operating a recycling platform.

7. The online platform management method based on multidimensional data AI artificial intelligence according to claim 4, characterized in that, The recycling platform refurbishes and processes used mobile phones, then entrusts them to retailers for sale. Users will see the recycling platform's origin and verification information on the retailer's platform. When the user's benefit from buying a new phone outweighs the benefit from buying a used phone... Users can buy new phones from retailers; when the utility of buying a used phone is greater than the utility of buying a new phone, the market demand for new phones will... When the utility of buying a used mobile phone is greater than the utility of buying a new mobile phone, Users buy used mobile phones on self-operated recycling platforms, increasing the market demand for new mobile phones. Second-hand mobile phones; among them, the market demand for second-hand mobile phones Retailer profits include profits from new phones and commissions from recycling platforms, while recycling platforms' profits include revenue from the sale of used phones and transfers to processors, minus input and output payments. （12） (13) Based on the retailer as the leader of the supply chain, the supply chain members conduct a two-stage Stackelberg game: In the first stage, the retailer determines the optimal retail price of the new mobile phone. In the second phase, the online recycling platform determines the optimal recycling price. Product promotion investment level and the selling price of used mobile phones ;in, 'f' represents the amount recycled, and 'f' represents the commission rate that the recycling platform needs to pay.

8. The online platform management method based on multidimensional data AI artificial intelligence according to claim 7, characterized in that, Under the strategy of self-operated sales of used mobile phones by recycling platforms, when the following conditions are met... and At that time, retailers and recycling platforms have an optimal decision: the optimal retail price of a new mobile phone. Best retail price for used mobile phones Optimal recycling price: Optimal level of investment in product promotion: 。 9. The online platform management method based on multi-dimensional data AI artificial intelligence according to claim 1, characterized in that, Generating user transaction orders based on the sales decision model and the model of the used mobile phone includes: obtaining the optimal decision by conducting sensitivity analysis based on the user's attention to product information and the reputation of the recycling platform, wherein the optimal decision includes the optimal used mobile phone selling price and the optimal new mobile phone retail price; generating user transaction orders based on the optimal used mobile phone selling price and the model of the used mobile phone.

10. An online platform management system based on multi-dimensional data AI, characterized in that, The online platform management method based on multidimensional data AI as described in claims 1-9 includes: a data acquisition unit, used to acquire multiple appearance images of a used mobile phone uploaded by a user to an online network platform and user registration information, and to perform image preprocessing on the appearance images to obtain image data, wherein the image preprocessing includes data cleaning, data augmentation, and image normalization; a feature extraction unit, used to extract image feature information from the image data using a MobileNetv3 network, and to perform similarity calculation in a preset vector library based on the image feature information to determine the model of the used mobile phone; a model building unit, used to build an IoT recycling model based on user registration information and the model of the used mobile phone, and to build a sales decision model for the used mobile phone based on user product attention and the IoT recycling model; and an order generation unit, used to generate a user's transaction order based on the sales decision model and the model of the used mobile phone.

Images